DDDA

Secondary data from non-centralized data source

Go to Data documentation

Danish Diabetes Database for Adults (The Danish Diabetes Database for Adults (DDDA)) (also known as Dansk Voksen Diabetes Database (DVDD) or Det Nationale Indikatorprojekt (NIP) Diabetes) is a no longer active database from regionernes kliniske kvalitetsudviklingsprogram (Regionernes Kliniske Kvalitetsudviklingsprogram (RKKP)). It was closed by 30 June 2022. It has now been replaced by Dansk Diabetes Database (DDD/DDiD), which has been active since 1 July 2022. Dansk Diabetes Database (DDiD) currently mainly uses registry data, however, as additional data are still being recorded in the clinic, it is worth checking the database in the future, as additional variables might be included (e.g. smoking, blood pressure, BMI, etc.).

Extracts from the online DDDA documentation (downloaded 07 September 2023) about the population and variables can be downloaded here:

And additional documents:

See also Jørgensen et al. (2016) for further information about the database.

Data

At Department of Clinical Epidemiology (DCE), we have data from DDDA in the period from May 2005 to January 2022 based on the n=10,241 individuals enrolled in DD2 per January 2022. As the database is now closed, this is the final data set and we will not get any updates of it.

In the latest version of the DDDA data, around 80% (n=8,512) of the individuals in DD2 had at least one record in DDDA. The records in DDDA are uniquely defined based on the CPR-number and the variable status_dato which is the date that defines the data entry. There is a median of 3 (IQR 2-5) records per individual, 85% had at least one record prior to enrollment in DD2, and there was a median of -56 days (IQR -413-294) from the enrollment date to the nearest status_dato (negative value meaning that status_dato was before enrollment in DD2). The status_dato is not necessarily the date of the examination/results/visit. Many of the variables thus have an associated date variable that is different from status_dato.

When using data from DDDA, it is important to keep in mind the missing data. Data may be missing either because it was not reported or because the individual is simply not registered in DDDA. Thus, when reporting data from DDDA, e.g., in a baseline table, remember to consider what “100%” should represent, as the interpretation of the result varies; is it x% of the entire study population or x% of the study population also registered in DDDA.

In addition, the selection and timing of data should be considered. Individuals registered in DDDA fulfill the criteria for this database (see population documentation above) which might induce some selection bias. Individuals in DDDA might not be as newly diagnosed as individuals in DD2. The records in DDDA are defined based on status_dato, which is “den dato, hvor der årligt, i relation til databasen, gøres status over hvornår personer senest har fået foretaget forskellige relevante undersøgelser” (see website). The dates listed for each of the examinations might be long after (or before) enrollment in DD2, and for each scientific study question, it is thus crucial to consider in which time period data can be considered valid. An example could be to only include measurements in the period from 5 years prior to enrollment to 6 months after enrollment and then among these, include the one closest to the enrollment date as a “baseline” value. This depends on the specific study.


Data documentation

DDDA.sas7bdat

Format (var x obs) Id variables Unique key Important dates
Wide (75 x 30,806) CPR CPR*status_dato status_dato, specific date variables

N=8,512 distinct CPR numbers with non-missing data are included in the data, i.e., individuals with data in DDDA. A total of N=10,219 individuals are included (merged with DD2core). A record is unique based on CPR and status_dato, however, for many of the variables, it is worth also considering the corresponding var_dato.

Illustration of the overall data structure. The dataset is in wide format (75 variables × 30,806 rows), with CPR*status_dato as the unique key.
Row CPR status_dato Var Var_dato
1 CPR1 status_dato1.1 num. Var_dato1.1
2 CPR1 status_dato1.2 num. Var_dato1.2
3 CPR1 status_dato1.3 num. Var_dato1.2
10 CPR1 status_dato1.10 (missing, no record) (missing, no record)
11 CPR2 status_dato2.1 num. Var_dato2.1
12 CPR2 status_dato2.2 num. Var_dato2.2
30,806 CPR10219 (missing, not in DDDA) (missing, not in DDDA) (missing, not in DDDA)

Variables from previous versions

In the current DDDA data, some variables from the database were not delivered. It was decided that the best suitable solution was to include these variables from older versions of data (and thus missing for some new individuals).

  • From the 2021 data: shared_care and Plasmakreatinin_operator

  • From the 2018 data (also missing in the 2021 delivery): diag_dato, HbA1c_kode, HDLcholesterol, and HDLcholesterol_undersoegelse_kod. These variables are included with a “_2018” postfix.

References

Jørgensen ME, Kristensen JK, Reventlov Husted G, Cerqueira C, Rossing P. The danish adult diabetes registry. Clin Epidemiol. 2016;8:429–34.